getzep/building-with-zep
Build and configure applications that use Zep — agent memory built on temporal Context Graphs. Bundles the building-with-zep, tuning-zep-ingestion, and migrating-zep-v3-to-v4 skills and the Zep documentation MCP server.
Guide for building, designing, reviewing, evaluating, and troubleshooting applications that use Zep, the unified context layer for enterprise data. Zep combines business data, documents, and conversations into governed temporal Context Graphs for three solutions: enterprise context graphs, agent memory, and customer and account context. Use whenever you write or design code that integrates Zep: giving an agent context or long-term memory, building a shared Context Graph, ingesting chat, business, document, or JSON data, retrieving a Context Block or searching the graph, choosing between user graphs and shared graphs, defining an ontology or custom instructions, applying access policies, or evaluating and tuning Zep. Triggers on "implement Zep for my agent", "set up a Zep ontology", or "why is Zep not returning the right context". For ingestion code or tuning, also use the tuning-zep-ingestion skill. Do not trigger when an end user asks an agent to remember or look up something in its own memory.
Procedure for a coding agent that migrates an application from the Zep V3 API and SDKs to V4. The procedure reads the V4 migration guide, finds each V3 call in the codebase, maps it to its V4 replacement, moves the application to Zep-generated UUIDs in the documented order, and verifies each step. Use when a user asks to "migrate to Zep v4", "upgrade zep-cloud to v4", "port my Zep code from v3", or "what breaks when I move to Zep v4". Use together with the building-with-zep skill from the same plugin. Activate that skill first if it is not active yet.
Procedure for an agent that must turn a dataset (database records, documents, email, chat messages, meeting transcripts, event logs, or a mix) into a domain-specific Zep Context Graph, with or without a gold question set. The procedure splits the data into a deterministic import and narrative episodes, shapes each episode, designs the ontology and the custom instructions, runs a versioned ingestion on a new graph, measures the graph shape, reads the debug logs and the Ingestion Traces, and changes one variable per run. Use when a user asks to "ingest my data into Zep", "improve extraction quality", "why does my graph have duplicate or untyped entities", "tune the ontology or the instructions", or "evaluate Zep on my dataset". Use together with the building-with-zep skill from the same plugin. Activate that skill first if it is not active yet.